First-generation products in all technology markets are good for proof-of-concept and demonstration of capabilities, but rarely do they make it from the engineer's bench to the mass market. And sometimes, even second-generation systems, while much improved, cannot achieve the proper balance of price, features and functionality.
In the development of the local multipoint distribution services market, the history is the same. When the first-generation systems were in trial, engineers were justifiably excited when they successfully transported data from one point to another at frequencies previously meant for high-cost defense applications. In order to achieve the reliability in the field required by commercial systems, solid-state components were needed. But at that time, the use of gallium arsenide (GaAs) as a substrate for chip design was neither ready for commercial use nor cost effective. GaAs was preferred over silicon because of the extra stability needed to support the high frequencies at which LMDS systems operate.
Besides the difficulty of making the wafers, one of the main reasons for the high cost of the GaAs was the low yield once the chips were etched onto the wafer.
At that time, it was normal to produce only a single usable chip from each wafer even though a dozen could fit in the space. The result, of course, was that the GaAs components came at a very high cost and mass production of components was nearly impossible.
Other challenges faced during this time were at the system level. Each radio required many hours of hand adjustment by experienced technicians to meet spectrum mask requirements. Compounding the challenge for the early engineers was that they were required to innovate on the "digital" side of the system as well. The demands for broadband were just coming to the forefront, and these new digital LMDS systems would need to not only operate at challenging high frequencies, but also provide access services at speeds never before supported by commercial systems.
First-generation point-to-multipoint (PMP) systems were based on frequency-division duplexing/frequency-division multiple access (FDD/FDMA) airlink technologies, which employed a single inefficient, although robust, modulation scheme, and limited capacity per channel. For decades, the combination of FDD/FDMA has been used for analog point-to-point (PTP) systems with one fixed channel dedicated for upstream traffic and an additional fixed channel for down-stream traffic. This architecture is sufficient for voice back haul where the symmetry is 1:1, and when there are no bursts of data traffic in the link. However, when deployed in an access network where data is carried, FDD/FDMA proved to be grossly inefficient and an expensive proposition for LMDS carriers.
Another area of technology development required for mainstream acceptance of PMP systems is the type and sophistication of the digital modulation schemes used to encode information for transmission. Given the large cost challenges to be overcome, these systems were primarily designed to provide the maximum amount of coverage, and therefore employed a robust but inefficient modulation scheme such as quadrature phase-shift keying (QPSK).
Ultimately, the use of expensive components designed for traditional speeds and incorporating inflexible technology to deliver basic services, first-generation systems saw limited deployments worldwide.
To overcome the significant technology challenges remaining with first-generation systems, in addition to the needed innovations in RF components, LMDS system designers set their sights on an enabling technology. "Burst modems" would be necessary to provide the core functionality that PMP promised, delivering instantaneous bandwidth on demand to various users over a geographic area on a burst-by-burst basis.
Second-generation systems that incorporated burst modems with time-division multiple-access (TDMA) algorithms were an initial attempt to allow flexibility in bandwidth allocations.
To increase capacity in second-generation systems, even at the expense of compromising range, more efficient modulation schemes of quadrature-amplitude modulation 16 (QAM16) and QAM64 emerged. However, there were coverage limitations with both of these new schemes.
While the more complex modulation schemes provided more bandwidth, implementing QPSK, QAM16 and QAM64 had to be done separately, channel by channel. Therefore, if four channels were needed for full coverage in a basestation, 12 channels then would be needed for the same coverage if all three modulation schemes were deployed.